Design: Choosing a camera system

22 Sep Design: Choosing a camera system

Tameem Antoniades

Hello, I am currently preparing for my first ever appearance at DICE Europe. It’s basically a conference for grown-up big-wigs in the games industry so I’m quite nervous about doing it. They originally invited Nina Kristensen, our Chief Development Ninja and co-founder, but she put me forward for it, despite my pleadings and protestations.

Anyway, despite that, I wanted to put something out there for you as Monday’s are becoming a regular thing in terms of releasing some info.  Melina is away on holiday so we don’t have a juicy video for you this time. Instead, I’ve put together a little design blog about choosing a camera system for Hellblade.  I think it’s worth reading if you are interested in nuts and bolts game design.  Otherwise, I’m afraid you may find it rather dry.

Choosing a camera view is one of the first and most important decisions to be made on a game.  It’s right up there with controls and character actions.  Is the game going to be 2D, isometric, first person or 3rd person?  Any system chosen will have massive ramifications on how you build the game, how you control it and how it will feel.  It may seem like the dog wagging the tail, but I feel it is smarter to design the game around your camera choice and not the other way around.

When I was working at Sony Europe about 15 years ago I had to design a 3rd person camera system for a game I was working on.  I analysed over 40 games, documented what worked and what didn’t.  It is much cheaper to learn from other games, both good and bad than to endlessly prototype what you think is right.

People may reasonably suppose that one of the most important skills of a designer is coming up with good ideas.  Personally I would put that way down the list of important skills.  Ideas are cheap, everyone has them, but few can articulate them, break them down, implement them and make them feel good.  So one skill I would put very near the top is “analysis”.

I reverse engineered and tested every camera attribute from the angles, the field of view, transitions, zoom behaviours, interaction with environment, controls and so on.  I then proposed several options before going deeper into analysis and choosing one proposal that I thought might work.  It took over a week and resulted in a 40-page tome but I learnt enough about cameras to last me a lifetime.  I still have that knowedge fresh in my mind when considering camera systems for Hellblade.

So based on what I knew already, I wrote a (thankfully) more brief analysis on our internal wiki.  It wasn’t written for anyone in particular, more to help my own thought process.  It’s sort of a hang-over from my computer science days when we were forced to do systems analysis.

I went into it this assuming that we would have a regular 3rd person camera like in DmC but the process of analysis lead me to a place I wasn’t expecting.  Here it is:

 Camera System Analysis

Objective

I cannot stress this enough: the gameplay style and levels we create are wholly dependent on the camera system we decide upon.

The camera system should support clear combat awareness and a feeling of immersion. There are broadly speaking two camera systems to choose from: a 3rd person chase camera and a fixed-rail camera system plus many hybrid systems in-between.

On all our projects other than KFC, cameras were a full-time job for a programmer plus considerable work for design. As we do not have a spare programmer and only one designer for Hellblade, the camera solution we employ must be one that does not require a dedicated programmer or onerous designer time to implement.

Another consideration is that once a camera system was implemented on our previous projects, I had to take directorial control over the camera placement, framing and editing throughout the game. This took an enormous amount of time and focus away from other areas of the game and I would like to avoid this going forward.

Chase Camera System

A single chase camera system that intelligently deals with enemies and environment. Examples include DMC, RE4, Last of Us, Batman etc

PROS

  • Only one camera is built for the entire game.
  • Player has a high level of control over the camera.
  • Framing of the character can vary but is relatively close, helping you connect with the character.
  • Exploration of level is freely attainable with this kind of camera
  • Supports varied gameplay modes from fighting to exploration to stealth to shooting

 

CONS

  • Development can take a very long time to manage every eventuality
  • The player is typically required to control the camera system to play effectively (not an issue for hardcore players)
  • Vista shots that show off the environment cannot be done with this system
  • Staging of level is difficult as we don’t know where the camera is looking at any point in time
  • Camera can get very messy during combat

 

Examples

 

 

 

 

Fixed Rail Camera System

Fixed camera are preset in maya or unreal to give a fixed and directed view of the action. ICO, God of War, Kung Fu Chaos, Heavenly Sword.

PROS

  • Very specific control over what is framed in the scene giving a very cinematic feel
  • User never has to worry about the camera controls
  • Action can be directed very precisely to the camera

 

CONS

  • Every part of every level has to be designed for the camera to make sure camera doesn’t swing, gimble-lock, get obscured etc
  • Exploration, stealth and so on is severely limited
  • Transitions from camera to camera or branching paths can be tricky to make feel right.
  • A huge amount of camera tweaking throughout development is needed on a scene by scene basis
  • Creating intimate camera shots can be a big challenge.
  • Suits very linear level design as backtracking involves running into the camera

 

Examples

 

 

 

 

 

Hybrid Fixed/Chase Camera Systems

A combination of fixed and chase cameras. Usually split to be chase for combat and fixed for traversal. Examples include DMC4, Enslaved.

PROS /CONS

  • The best of both worlds or the worst of both worlds depending on your implementation
  • Requires the implementation of two camera systems and seamless blending between

 

Examples

 

 

 

Directed Chase Camera

This is a chase camera that points at the framing that you decide as a player but that can be overridden at any point using camera controls. Examples: I think uncharted 2 used this, and Mario 64 in places.

PROS /CONS

  • The best of both worlds or the worst of both worlds depending on your implementation
  • Requires the implementation of two camera systems and seamless blending between

 

Examples

 

 

Vector Field Camera

This is more of an idea I’ve had for a while rather than a proven solution.  I imagine a 3D vector field covering all playable areas of a level that contain position, direction, character offset and fov parameters. The actual camera would interpolate between these vectors. You could set the vectors by flying the free cam and positioning the hero in-game until you have enough vectors to describe the whole space.

UE4 supports vector fields so I don’t know if that is something that may be repurposed to control cameras. Perhaps it’s worth investigating.

PROS /CONS

  • Would support larger explorable areas unconstrained by spline paths.
  • Would require lots of prototyping and implementation

 

Over-the-Shoulder Camera

This is an over the shoulder view with controls more in common with first person shooters than traditional third person views. It requires that controls are mapped to support this mode and suits shoulder buttons for man actions as you would have to release the right stick to press a button. Examples include Dead Space and RE4.

 

Examples

 

 

 

 

PROS 

  • Very immersive. you feel close to your character
  • Fewer collision issues as the closeness limits obstruction
  • Simpler implementation as it doesn’t require as much “assistance” as chase cameras nor the arduous set up of fixed cameras.
  • Supports very cinematic transitions (like Enslaved’s dutching and Dead Space’s moments)

CONS 

  • Turning to face enemies is clunky as controls must be character-relative rather than screen-relative.
  • Visibility of surrounding enemies is restricted so gameplay should limit these to a few.
  • Face buttons are less accessible which could present serious problems for combat.

 

Additional Notes:

A shoulder camera will have a radical effect on controls, level design and enemy make-up, AI and combat and will take us into largely uncharted ground.

1st Person Camera

There are few examples of this in melee games. There are simply too many problems with the narrow visibility and not being able to read the character attacks and reactions on screen clearly. Though it simplifies some aspects of development, namely the fact that you don’t need to see/animate your character fully, there are far too many risks without clear tangible benefits to the player for this to be a serious consideration.

 Investigate

  • There is a default chase camera in UE4. We should put this to the test in a test map to see its potential and limitations.
  • Can we re-use camera systems from any previous projects?
  • What about a shoulder camera? We should protoype and explore this?
  • Can vector fields or SHL systems be hooked up to the camera in UE4? Can this be done without code support to test the idea? What tools are available in UE4?
  • The DmC chase camera was very fully featured but i imagine it was massively over developed and unusable in a new UE4 code base? Confirm
  • Kung Fu Chaos had a maya-exported camera. This was fully controlled by the artists in maya. For this to work, the levels have to be extremely linear.  Discount because workflow should remain in UE4 and limitations are too onerous.

 

Considerations

  • Hardcore players don’t mind manual camera control but this comes with caveats that suggests lots of camera development:
    • The camera must follow the action invisibly. e.g. when fighting an enemy/enemies, it should track them
    • The camera must zoom in and out intelligently depending on number of surrounding enemies
    • The camera must never swing rapidly or otherwise cause sickness
    • The camera must deal with obstruction effectively
    • Wide open spaces, corridors, slopes should have different framing and positioning
  • Our code resource is severely limited so the only obvious way we could implement an intelligent chase camera is to port it from one of our other games (DmC, Enslaved).  Based on initial investigations, this is not feasible without a lot of code refactoring.
  • Examples of fixed-camera games are few and far between. It seems to have fallen out of favour in the name of giving players the freedom to explore. God of War is probably the last example of such a camera system. It is risky to bet on it.
  • Shoulder cameras are used for exploration/shooting games and I can’t think of good examples of melee games that use this viewpoint other than God Hand which great combat controls but had clunky movement controls. Doing a melee game with this viewpoint is a risk but could perhaps present an opportunity to innovate.

 

 Conclusion

  • We should pursue and prototype the shoulder camera as the first port of call.
  • The shoulder cam has a far reaching consequences (effectively, you’re making a very different game from DmC) but presents several opportunities:
    • The viewpoint is unusual for melee games. All existing examples are from PC games/mods from devs that are traditionally 1st person devs.
    • Translating the depth of combat of DmC and fighting games into this viewpoint will be innovative and could present a progressive direction for melee.
    • The sense of immersion is directly comparable with the shift from RE1 to RE4 in both gameplay and cinematic moments.
    • Traversal in this viewpoint can be more scripted and canned like in RE4 rather than platform based like in DmC/GoW.
  • There are several limitations to bear in mind:
    • Large open arenas are more suited to 3rd person camera views like in DmC whereas shoulder view suits more corridor based environments
    • Shoulder-based FOV limits surroundings so having lots of enemies surround you is out. Better to focus on fewer tougher enemies that approach from in front.
    • Enemy AI would have to respect the players’ limited viewpoint.
    • Controls are may have to be dictated by shoulder buttons as face buttons are difficult to reach (this may kill the idea of a shoulder cam – needs to be prototyped)
    • Animation pace and realism will matter more as you are closer to the character.

So summing up, we simply don’t have the resources to implemented most of our usual options.  An over-the-shoulder camera is a more reasonable option from a resource point of view but leaves many questions unanswered when it comes to combat.  This became the top item to prototype.

At the time this analysis was written, we didn’t have any available code time on the project.  Luckily Alexander Litinov had just joined the team as THE designer and comes from a programming background.  He prototyped the camera system in Unreal using its Blueprint system.  He had done enough for us to play around and experiment with different control options with different numbers of enemies.  I was delighted that it seemed perfectly natural to use face buttons in this camera view. It meant we could create a hardcore fighting set-up within an unusually intimate camera view.  In a sense it felt a little more like a one-on-one fighter than a 3rd person one-on-many masher which is no bad thing.

So unless something unforeseen derails us, an over-the-shoulder style camera system it is.

We will be probably be covering our prototype phase in the next video diary so there should be more details then.  Stay tuned.

16 Comments
  • Bren Turner
    Posted at 14:31h, 22 September Reply

    A camera system similar to the Batman Arkham series would be nice, if you ask me.Also, the camera system used in Max Payne 3 can be a good reference too.

  • Henrik Mygren
    Posted at 15:19h, 22 September Reply

    Hello Mr Tameem are you gonna visit Scandinavia in search of insperations?

    I know Britain has alot of info already seeing as the country was occupied for well atleast a couple of hundred years atleast.

    But seeing scandinavia firsthand migth inspire further.

    Or maybe Vikings tv series,while alot is historical liberties most is correct.We where not all barbarians 😛 we where just farmers that wanted your lands to grow crops on ;P

    I guess Hellblade might not be that historical correct in the end anyway but insperations are good 😀

    I do like your setting anyway.

    About camera I do like cinematic camera up and close with characters however as you yourself said migh be hard if Hellblade shall picture epic grandscale battles its hard to fit.

    But taking enlslaved,DMC and Heavenly Blade in consideration I trust you will find what fits the game best as you have not disappointed me yet anyway.

    Good luck with Hellblade development.

    Also its a bit Ironic for what are you(if I might be blunt) Mr Tameem British/Indian man to come up with such a great setting of Hellblade..
    We Scandinavian tend to forget our own history and what great setting it could be in movies and games that others tend to make it for us instead.

  • Michael Shade
    Posted at 17:17h, 22 September Reply

    Fantastic post!

  • Zura
    Posted at 17:40h, 22 September Reply

    Very interesting!

  • John
    Posted at 18:27h, 22 September Reply

    I really like the idea of fewer thougher enemys. In my opinion too much enemys will make the fights less intensive. For example the fights in Assasin’s Creed, I started to like the series less and less because of the fighting system, where you can just counter one attack and the effortlessly kill everyone.
    However, the article was really interesting and I’m looking forward to watching the next video.
    Keep the great work up!

  • André Rogers
    Posted at 21:01h, 22 September Reply

    I agree with Bren Turner’s suggestion about the camera system used in the Batman Arkham games. The camera in these games tends to stick close to the back of the character almost reminiscent to a shoulder camera but pans out during certain action sequences to give a better field of view.

  • Henri Eberhard
    Posted at 22:04h, 23 September Reply

    the idea of a hard core fighting set up from a shoulder view is interesting. Can I ask though if it is really impossible to have multiple enemies with this set up? I understand that it is difficult for very obvious reasons but not impossible and I think that fights against groups of enemies play to the strength of the shoulder camera in regards to adding to a sense of claustrophobia and paranoia that will further immerse players in the combat e.g. resident evil 4, dead space (though not against nearly as many enemies in comparison with those games). the idea that you have fewer tougher enemies would be very interesting, and to have to manage a battle against a group with such a restricted camera would promote the importance of the players position in a fight and the observation of the enemies you are fighting.

    In thinking about this further I don’t think a similar style of game to to devil may cry or god of war can be accomplished with a shoulder camera. It will alter combat to approach greater realism. and although keeping combat to one on one fights will make things easier to keep to the hard core fighting game style, I think that it is ultimately a disadvantage for you because it will restrict the development of the games combat from becoming something unique that could otherwise come out of a game that has combat that is informed by the way the shoulder camera works.

  • Pablo Amaral
    Posted at 11:51h, 24 September Reply

    Very good post, last semester I’ve made a game and DMC was my biggest influence to produce the câmera for it. This post is very complete, since cameras are very difficult so it needs a lot of study, thank you.

  • Clint Oliver
    Posted at 15:48h, 24 September Reply

    Thank you for sharing with us, Tameem. I’m definitely intrigued how the team plans to explore fast and intensive melee combat with an over-the-shoulder view. One of my initial concerns is that the enemy design and character behaviors (especially that of the player character) would have to be very static. That is, there would not be much dashing around or quick movements performed by the player character, as the camera might make the visual appear jerky or clunky.

    I look forward to seeing how the team plans to overcome or diminish the cons of an over-the-shoulder system. Thank you for opening up the visibility of the team’s development processes. I’ll continue to follow the development and offer my thoughts and ideas when applicable.

    Have a great day!

  • Graham Raddings
    Posted at 08:26h, 25 September Reply

    I think over-the-shoulder camera based combat will prove itself (as was the case with Ryse) to be arduous and lumpen. It does not lend itself to frantic classic hack-and-slash based combat. I would love to see some footage of the early camera experiments to see how it played. The closeness would definitely make it more personal – would an implementation similar to the recent Tomb Raider work in this way? Mind you I suppose that lent itself to aiming and shooting (much like Uncharted) than the up-close and personal DMC approach. I would be interested to know which angle you are coming at this from (if you’ll pardon the pun). Is this game a point-to-point combat based hack and slash, or is there a deliberate strategy to fuse a more RE4 notion into the gameplay? Either way this is a great read and I do hope you’ll follow it up with some footage of the camera experiments.

  • Deividas
    Posted at 04:01h, 29 September Reply

    Hello,

    very interesting post. If the game is to have exploration and platforming I would suggest Batman Arkham camera as well, though if this is going to be an action game (I hope it will) I would say DMC camera or Ninja Gaiden camera would suite it perfectly. About lots of enemies on screen, in my opinion more than 10 enemies is a crowd, and that would be frustrating or the enemies would have to be really dumb. My honest opinion You have made Heavenly Sword and DMC, make the best of these game with new original story and we will have an awesome game. One thing though, ability to block is essential in my opinion. And for me smaller but more interactive levels (DMC) are better than gorgeous, but dead vistas (Ryse) also I don’t know about the prices and how hard is to optimize the game, but Havok physics but be cool as well. For example a level were tables, chairs, bottles etc. are standing still and as you throw an opponent for example, everything just brakes and flies everywhere. By the way I really like what you did in DMC regarding weapon switching to defeat certain type of enemies. Also thank you for making this game

  • Rustam
    Posted at 13:23h, 03 October Reply

    It would be nice to have battles like in 2008 PoP but with more harcore approach, blood and gore and such stuff, with your concept of over-the-shoulder camera it sounds interesting, but such a tiny team cannot, i assume, make as much animations for combos and QTE/Highlight-kill camera like we had in PoP, which is rather sad, because I liked improvements in cobat in DmC, there we hade so many combos and stuff compared to previous Ninja’s games, where all combat was down to mashing buttons. I hope you can create such a duel-type combat with many options and animations, but, on other hand, that you do not lose all the charm that world and characters meant for your games, your unusual approach and style.

  • zagham
    Posted at 08:00h, 05 October Reply

    Have an option of over the shoulder and directed chase camera

  • zagham
    Posted at 08:21h, 05 October Reply

    Have a different male character added in the near start.
    Or have a nale character in legendary mode for when game ends

  • guest
    Posted at 18:16h, 13 October Reply

    How about a camera system like the one used in the (Dark/Demon) Souls series ? It’s great for combat

  • Niccelson
    Posted at 09:30h, 17 October Reply

    I can’t really comment on the camera choice without seeing any resulting gameplay, but I will say this: PLEASE INCLUDE A LOCK-ON BUTTON! The lack of this feature was the only thing that soured DmC for me, as without it long range attacks constantly miss the target and enemy dodges can put them out of camera view entirely.

Post A Comment